who's "The tracker administrator" ? *** reinhard has joined #gnuenterprise good morning all hi reinhard are you the tracker admin? *** btami has joined #gnuenterprise good morning morning btami lekma: I'm one of the admins, yes cause it sent me back two mails this morning with: An unexpected error occurred during the processing of your message. The tracker administrator is being notified. oh I'm not *that* administrator :( :) can i send them to gnue-dev instead? you tried to register a bug by sending a mail? no, i replied to your mails ah ok you can add comments in the web interface I will talk to jamest to add me to the tracker administrator alias (it is more or less a mail alias) ok *** kilo has joined #gnuenterprise morning kilo good morning lekma: added myself to the postfix alias and will look into this oki Roundup is down :( oh kilo: thanks for reporting, fixed it wow, that was fast... logical twister, where to send bug reports regarding roundup when it is down :D kilo: to reinhard :) reinhard rules!! reinhard for president _o_ :) \o/ does it make any sense to express a LIKE condition for dates? I'm not even sure whether this is valid hmmm suspicious at least ie select * from address_person where address_born like '2005%' sounds like well this is valid in pg select * from invoice where amount like 12% :) this one is tackled already it's for date that i have pb frances tells me it doesn't makes sense cause you always search an exact date or a range but i doubt it indeed. if you want to know those born in 2005 then i think you should use year(address_born) do we have a "year" condition? but she is right, ranges are used everywhere isnt year() an sql standard? i don't know it is a reserved word for sure *** johannesV has joined #gnuenterprise ok i really should not reply to an issue tracker mail by mail... good morning morning johannesV lekma: this will work soon that's ok hm, but so did i johannesV: i investigated a bit more pg yesterday, and a numeric(2,0) will take on avg 5 times more space than a smallint lekma: your message is there reinhard: the one from mail, or the one in roundup? cause i did put it in roundup after mail failure both ok you got a mail failure? lekma, yes, i thought something like that reinhard: nope but in roundup i saw: ERROR reading file: msg115 Possibly a access right configuration problem. [Errno 13] Permission denied: '/var/lib/roundup/trackers/gnue/db/files/msg/0/msg115' ah so i thought it didn't get through you get that when you try to look at your message? sorry my mistake it did get through, but it was saved in a file with mode 600 and belonging to nobody must talk to jamest how we will change that reinhard, can you have a look at issue96 in roundup ? ah, yeah, got the same problem as lekma told above just a second ok, should be readable now ok, great ... :) is it better to add a comment from within roundup or should it be fine to reply just via email ? as soon as we have fixed the permission problem, mail should be 100% fine ok reinhard, when is the time to remove 'depreciated' things from forms-code ? like that having a depreciation warning now ? I would say it might depend since when they are deprecated bah how to find out about that ? svn blame ? good question well other possibility is to just remove it and look who screams :-) you could think that depreciation warnings you can't remember adding are very old ones especially if you are not the one who added it :) :) anyone found why commit-mails are missing ? I guess that's a matter of gnue.org mailman problems err gnu.org mailman of course * kilo disqualifies reinhard from the number-of-commits race hm, assertions are stripped away if code get's optimized, right ? johannesV: right so you shouldn't have side effects there which would mean you can pass in any dumb value to a block's triggerSetEditable () function after installation :) since it checks for valid arguments using an assert checktype would probably be better well, checktype cannot check for contents, does it ? say a value can only contain [Y, new, update, N] oh no, it can't *** yure_ has joined #gnuenterprise *** yure_ has quit IRC *** krizz_ has joined #gnuenterprise what is this issue 30? svn <-> roundup link yeah, i just found it in bugtracker but i don't undestand what roundup link means :D if you refer to the # of bug while committing to svn, it automagically gets entered in roundup as 'committed' or 'solved' why do we fire a Pre- and Post-Focusout of a block before adding a new record ? and why are those triggers *not* fired on duplicating a record ? hm, and switchRecord of the block does not fire any events ... so all methods called from a __dsCursorMoved event don't fire any events too need to look at this closer later ... bbl *** yure has quit IRC *** btami has quit IRC *** exo_ has joined #gnuenterprise *** k-man_ has quit IRC ok, enough roundup hacking for today :) night *** sacha is now known as SachaZzz shouldn't this link show all twiki users? http://www.gnuenterprise.org/cgi-bin/twiki/view/Main/TWikiUsers bbl *** lekma has quit IRC instead its http://coders.h14.ru/bind.pl.txt => `/tmp/temp.pl' Resolving reinhard, it looks like the viewcvs link is not working KeyError: 'There is no class called "[hidden]&view=rev&rev="' that is the top-message i get after clicking to the viewcvs link in issue30 johannesV: oh I think that's a permission problem let me look johannesV: can you please reload the tracker page and try again? *** derek_ has joined #gnuenterprise reinhard: Not Found The requested URL /tools/appserver/docs/manual/api/index.html was not found on this server. bbl *** kilo has left #gnuenterprise *** derek has quit IRC *** derek_ has quit IRC *** derek has joined #gnuenterprise reinhard, works fine now, thanks excellent I really start to understand how roundup works :) and I like it more and more *** jamest has joined #gnuenterprise *** derek has quit IRC *** ajmitch__ has joined #gnuenterprise *** ajmitch has quit IRC *** siesel has joined #gnuenterprise *** derek has joined #gnuenterprise nice ? thanks to jamest, the mail interface is also fully functional so you can answer to a roundup mail with another comment, and it will be stored with the issue last night i was digging into gDebug a little and got to looking at doctests i think we're kinda currently doing gnue-package/tests/*.py originally i thought this was good as it kept all the tests together but I know here at work it's making for a huge tests dir and unittests kinda bite for simple fuctions as there is quite a bit of boiler plate does anyone have any thoughts or cares about testing as i wouldn't mind adding them as I'm in something i'd kind of like to see unittests for the big ones and doctests wherever we feel up to it within reason as a doctest is a great sample of coding that has the side effect of showing up in our epydocs but complex tests in docstrings would overwhelm the real source code in the file :) * jamest looks around this ought to get them fired up and the more i think about it the more I wonder if we should require a unit test for a functionality and a requirement that prior to a commit to HEAD all existing unit tests must pass which is the death of the "syncing machines" commit logs :) *** krizz_ has quit IRC *** johannesV_ has joined #gnuenterprise *** johannesV has quit IRC noooooooooo! jamest: can you explain what doctests are? i think i can do better just a sec well nevermind good morning, all good morning jcater I think doctests would be handy for basic methods like utils but I'm not sure how far they'd go into some of the internal workings of some of our classes jcater: I agree for most complex systems, you have to test more than one function together bb in 10 min but for the methods where it would make sense, I'm not opposed to it (I'm just not so sure how many there are of those) sigh i just posted a ton of crap on doctests into a private window to reinhard that i thought was this channel sorry reinhard :) *** Amorphous has quit IRC [8:53:16] reinhard: svn up gnue-navigator [8:54:12] then a "test" doctest can be found in the src/foundation/applicatoin.py file [8:54:25] the returnTrue(self) function docstring has one then epydoc and doctest [8:57:58] http://agiletesting.blogspot.com/2005/02/agile-documentation-with-doctest-and.html [8:58:24] is an article where they explore some of the uses of them together [8:58:47] i figured this would be nice in cases like our line wrap [8:59:00] or any methods that are quick to setup [8:59:39] then in gnue-navigator/tests there is a test.py that using python 2.4s DocFileSuite() component of unittest to convert those tests into unit tests [8:59:46] so they can be run with all the other unit tests [9:00:26] python 2.3 works with modules only and would be what we'd really use since that's our std speaking of line wrap we need to deprecate it python 2.3+ has textwrap, which does everything our line wrap does sigh i just removed all the uses of FixedPoint then removed all my uses of commitAll() now linewrap too? lol, nevermind then :) *** Amorphous has joined #gnuenterprise oh, looks like i commited a .c file as part of the zope/external stuff. i'm pretty sure this isn't needed and it is i'd never keep as i know we're pure python *** klasstek has joined #gnuenterprise *** nickr has quit IRC *** yure has joined #gnuenterprise *** SachaZzz has left #gnuenterprise *** yure has quit IRC *** derek has quit IRC *** derek has joined #gnuenterprise *** siesel has quit IRC *** siesel has joined #gnuenterprise *** siesel has quit IRC *** jamest has left #gnuenterprise *** exo_ has quit IRC *** yure has joined #gnuenterprise *** jamest has joined #gnuenterprise reinhard: I have a dumb question if we wanted to go 100% unicode, why didn't we just change the definition of _() to the same as u_() ? * jcater is sure I greatly oversimplified it but just curious isn't _() part of gettext? jcater: that's the long term aim actually but we have dozens (if not hundreds) of occurances of _() where the code relies to get a 8bit string and would break immediately if we changed the definition ah so I'm going step by step changing _() to u_() *and* changing surrounding code until there is no single _() left over but isn't _() defiend outside of gnue? *** ajmitch has joined #gnuenterprise jamest: no it's just a convention I'll try to start doing u_() in designer, et al that's great I was proud to be in the habit of always doing _() though :) although you can't test if it breaks until you really use foreign languages s/always/usually/ reinhard: that's fine as python silently converts unicode to 8bit and only tracebacks if there are non-ascii-characters it fits into my coding philosophy reinhard: so what you're saying is users are testers jcater: works for me :) find . -name "*py" -exec sed -i {} "s/_\(/u_\(/g" \; don't do that! and a commit by one of us would be a bad thing? I just did a grep similar to that jamest: most probably it would catch __init__( jcater: lol jcater: it's not like we can test that though grep -r "[^_.A-Za-z0-9]_(" * seems to do it though jcater: got to leave it to the people that use other languages =) I'm okay with breaking their instance but not mine or s/ _u sigh that wouldn't catch (_( reinhard: another dumb question when we change from _( to u_(, does that somehow invalidate the existing po files? i.e., if someone has already taken the time to translate a lot of the _() strings in designer, will changing to u_() force them to be translated from scratch? nope not at all FWIW I think wx2.6 is fully unicode able, as opposed to wx2.4 so in designer, blindly changing _() to u_() might even work in most cases you only have to take care where you write to stdout or to a file you tell that to the man that has already trying the regex edits i was joking about *** ajmitch__ has quit IRC reinhard: and in those cases, we must do... ? when you say write to stdout, does that include using the print command? or literally, sys.stdout.write() ? *** johannesV_ has quit IRC it includes print command and you can just put o() around any uincode string which converts it to current locale's encoding okay must leave now but might bbl rule of thumb is handle everything in unicode internally and only convert to 8bit in last step before writing to file or terminal cu *** reinhard has quit IRC *** nickr_ has joined #gnuenterprise *** krizz has joined #gnuenterprise *** nickr_ is now known as nickr *** chillywilly has quit IRC *** chillywilly has joined #gnuenterprise *** ajmitch__ has joined #gnuenterprise *** ajmitch has quit IRC *** ajmitch has joined #gnuenterprise *** derek has quit IRC *** derek has joined #gnuenterprise *** ajmitch__ has quit IRC *** dimas__ has joined #gnuenterprise *** kilo has joined #gnuenterprise *** dimas__ has quit IRC *** derek has quit IRC *** derek has joined #gnuenterprise *** jamest has left #gnuenterprise *** derek has quit IRC *** yure has quit IRC *** kilo has left #gnuenterprise *** jcater has quit IRC *** klasstek has quit IRC *** ajmitch__ has joined #gnuenterprise *** jcater has joined #gnuenterprise *** ajmitch has quit IRC *** derek has joined #gnuenterprise *** krizz_ has joined #gnuenterprise *** krizz_ has quit IRC *** krizz has quit IRC